Commit Graph

1444 Commits (49bdcef648d5aa127e40b015217aa11ac0f53973)

Author SHA1 Message Date
Michel Juillard 956b42fdff corrected bugs in shock_decomposition 2010-04-27 19:50:42 +02:00
Marco Ratto a6ac64e5db small bug fix 2010-04-20 10:37:08 +02:00
Marco Ratto 47037687d5 fixed persistent variable initialization 2010-04-19 16:40:28 +02:00
Michel Juillard 4bde063669 Kalman filters: fixing initialization of oldK to Inf instead of 0 2010-04-16 12:20:57 +02:00
Sébastien Villemot 5b851c8c76 Merge remote branch 'ratto/master' 2010-04-15 16:13:38 +02:00
Michel Juillard 07748100ff adding to diagnoosticks check of rank of Jacobian of static model when steady state computation fails 2010-04-14 20:58:54 +02:00
Marco Ratto b6ad0a61ed bug fix for empty global variables 2010-04-14 17:29:12 +02:00
Marco Ratto 2d21729598 bug fix for non-existing remote sub-folders 2010-04-14 17:23:26 +02:00
Marco Ratto 47eac6ac88 1) copy metropolis files
2) bug fix with mh_drop
2010-04-14 17:20:29 +02:00
Marco Ratto 1fc0fbc2c7 Extended remote delete, to clean everything after completion of jobs on remote machines. 2010-04-14 17:19:55 +02:00
Marco Ratto 0e98d4266c 1) Several bug fixes for making the remote threads work properly under the Always/Open mode;
2) small cosmethics;
2010-04-14 17:19:21 +02:00
Marco Ratto f743f0375d small bug fix for remote slave processes; 2010-04-14 17:18:22 +02:00
Marco Ratto 5e7d9d39e5 1) removed redundant loop defining x;
2) bug fix for remote slave sessions;
2010-04-14 17:17:22 +02:00
Sébastien Villemot da74e53204 Preprocessor:
* give the possibility of calibrating measurement errors in the "shocks" blocks (only for observed endogenous variables)
* M_.H is now initialized in the preprocessor
* only one "varobs" statement is now accepted in a MOD file
2010-04-14 15:03:41 +02:00
Marco Ratto 0f06026c85 1) No exception for mode_compute=5 when computing hessian;
2) bug fix;
2010-04-08 17:04:08 +02:00
Sébastien Villemot a98a5106e1 In MATLAB >= 7.7, don't display a warning if we use deprecated interface to set seed of random number generators 2010-04-06 16:52:05 +02:00
Michel Juillard c2f7f0a555 computes now variance decomposition relative to the sum of the effects of individual shocks rather than aggregate variance.
When the aggregate variance differs from the shock of the sum of the effects of individual shocks by more than 0.01% a warning is displayed.

This behaviour is documented in the reference manual.
2010-04-03 11:27:49 +02:00
Michel Juillard f85049e9a6 removed oo_.smoother.integration_order
removed integration_order (d) from the output arguments of DsgeSmoother. It is still computed by a few Kalman smoother routines. I don't change them now as they need to be rewritten soon.
2010-04-03 08:00:22 +02:00
Michel Juillard c1b3627896 correction swz/c-code/Makefile: sbvar_estimation was left from a previous experiment
correcting swz/bin/.gitignore
2010-04-03 07:50:48 +02:00
Michel Juillard 6a2d2a7b46 smoother deals with all variables in M_.endo_names unless option SELECTED_VARIABLES_ONLY is used 2010-04-02 21:40:50 +02:00
Michel Juillard e54a505159 bug correction 2010-04-02 21:38:05 +02:00
Michel Juillard 2679e190e8 bug correction. It seems that struct2local can create problems with Matlab r2010a: if run_smoother exists as function name, 'if run_smoother' calls run_smoother rather than evaluate local variable run_smoother. 2010-04-02 21:37:43 +02:00
Stéphane Adjemian (Ithaque) aef46cf78e Fixed bug. The declared paths for the constrained variables have to be centered (Users declare constrained variables in *levels*,
imcforecasts centers the paths and adds the trend in the end).
(cherry picked from commit 4cb664a5089bc25bba7dfe63c2f0bdb743c4d5a7)
2010-04-01 15:41:36 +02:00
Sébastien Villemot 1dd9ca8e3f Activate warning backtrace even for Octave 2010-03-29 12:44:56 +02:00
Sébastien Villemot 7f5f99b43e stoch_simul: force order=1 with partial_information 2010-03-29 12:44:38 +02:00
Michel Juillard 307911b461 removing DOS end of lines 2010-03-25 12:28:48 +01:00
Michel Juillard e405f17c8a additional fixe to partial information:
- partial_information is a LOCAL option that must be repeated in all instructions
 - when VAROBS is empty, one assumes full information (all endogenous variables are observed)
 - allusion to SIMULATED variables is removed from the title of the tables
2010-03-25 11:20:34 +01:00
George Perendia 0be001d194 Tidying up Partial information extension to stoch_simul, PCL_Part_info_irf.m and PCL_Part_info_moments.m 2010-03-24 23:26:05 +00:00
Sébastien Villemot 667d5ac262 var_listindices.m: fix for Octave 2010-03-24 12:54:18 +01:00
Sébastien Villemot 6bb4ec4d9b Fixed bug in stoch_simul related to the variable list 2010-03-24 12:54:02 +01:00
Sébastien Villemot 60ad78b095 Partial information: fixed copyright notices 2010-03-24 12:21:02 +01:00
Sébastien Villemot 45efcd61dd Partial information: converted files to UNIX EOL convention 2010-03-24 12:14:41 +01:00
Michel Juillard a9ce1ef2b4 bug correction in partial_information 2010-03-24 10:35:30 +01:00
Michel Juillard 7668bc4a11 -new smoother function kalman/smoother/kalman_smoother.m
-fixing bugs in dynare_estimation_1.m
2010-03-23 18:17:15 +01:00
Michel Juillard b261eb0ba2 integration of partial information in stoch_simul (thanks to George, Joe and Paul) 2010-03-23 11:09:59 +01:00
Michel Juillard 436fdf97c6 modified varlist_indices 2010-03-23 06:38:39 +01:00
Michel Juillard c5bc424dd4 added new Partial Inforamtion code sent by G. Perendia 2010-03-23 05:47:13 +01:00
Michel Juillard 9b20ae958c added options filter_covariance filter_decomposition for new version of the smoother 2010-03-22 21:57:46 +01:00
Michel Juillard c963fb6c91 Merge branch 'master' of ssh://kirikou.dynare.org/srv/d_kirikou/git/dynare 2010-03-19 20:41:04 +01:00
Michel Juillard 40f1ab7548 computing smoother only for listed variables (start) 2010-03-19 20:40:49 +01:00
George Perendia 5efb7a7ddf added clause in dynare_estimation_1.m to avoid use of kalman smoother and diagnostics if options_.partial_information == 1 2010-03-19 19:17:38 +00:00
Michel Juillard 6d4b0ecab7 SWZ:
- reverting part of last change to handling of Qi and Ri in ComputingTasks.cc: r is the number of lags present in the restrictions
- fixing handling of Dirichlet option in mhm_input.dat. This fixes the strange Signal messages when running C executables.
- fix path in test_upper_cholesky.mod
2010-03-15 12:03:37 +01:00
Michel Juillard 890a7b8f0a SWZ: removed reference to old function 2010-03-15 10:10:22 +01:00
Sébastien Villemot afc94953e9 Add provisions for MATLAB 7.10 (R2010a) 2010-03-10 18:44:46 +01:00
Michel Juillard 5b1ebc821c SWZ: adding new tests and identification/exclusions.m 2010-03-10 08:40:52 +01:00
Michel Juillard 591e426110 - reversed ordering for lower and upper Cholesky
- added tests
- fixed various bugs
- modified initial values
Still unfinished
2010-03-10 08:32:51 +01:00
Michel Juillard f4b7840739 corrected bug for decompostion (nargout > 7) in missing_DiffuseKalmanSmoother3.m 2010-03-09 18:43:50 +01:00
Houtan Bastani 30ea396a08 Allow temporary terms to work with external functions 2010-03-09 11:21:14 +01:00
Marco Ratto 38640c4af8 use of numerical Jacobian to check analytic results. 2010-03-09 10:05:13 +01:00
Marco Ratto ed4fee93fb Utility for numerical derivatives in identification (to check analytic results) 2010-03-09 10:03:05 +01:00